Ian FREISLICH wrote:
> "Mark Smith" wrote:
>
>>>-----Original Message-----
>>>From: exim-users-bounces@???
>>>[mailto:exim-users-bounces@exim.org] On Behalf Of Ian FREISLICH
>>>Sent: 28 October 2005 09:10
>>>To: exim-users@???
>>>Subject: Re: [exim] Spam block idea:
>>>
>>>
>>>Drop in the rcpt ACL would just 5xx that recipient, but
>>>not the entire mail and the sending MTA (if it's an MTA) will
>>>retry later with the good recipients.
>>>
>>
>>That isn't the case. Drop immediately terminates the entire transaction,
>>before the data ACL is even reached.
>
>
> So lets examine the case:
>
> recepients 'a', 'b', 'c' and "fake" 'd'.
>
> HELO ...
> 220
> MAIL FROM:...
> 250
> RCPT TO:<a>
> 250
> RCPT TO:<d>
> 550
> Connection closed by foreign host.
>
> Does the remote side give up on recipients 'b' and 'c' or does it establish
> a new connection some time later and try them as well as 'a'?
>
> RFC2821 (3.9) implies that it will retry:
> SMTP clients that experience a connection close, reset, or
> other communications failure due to circumstances not under
> their control (in violation of the intent of this specification
> but sometimes unavoidable) SHOULD, to maintain the robustness
> of the mail system, treat the mail transaction as if a 451
> response had been received and act accordingly.
>
> Hence "remember that a fake address was a recipient and use deny
> or drop in your pre-data ACL". That way you kill the whole mail,
> not just one recipient.
>
> Ian
>
> --
> Ian Freislich
>
Interesting thread. Might I ask, what is the likely effect (if any)
when one's Exim is configured to neither advertise nor accept 'pipelining'?
Bill Hacker
